MOL future plans

[mol-general@lists.maconlinux.org]

On Sat, Apr 06, 2002 at 12:59:39PM -0700, Keary Suska wrote:
> I would rank more complete docs higher than man pages. Right now, the docs
> suffice mostly to just get MOL working, which doesn't take much, but to make
> it useful, certain critical components are missing. 

> Namely: #1 Networking.

Yes, this is poorly documented. A complete description of everything 
together with examples are needed.

> The options aren't spelled out or the fact that TAP/TUN interfaces are
> considered deprecated in 2.4.x kernels. 

Ethertap is deprecated, but not the TUN/TAP bridge (or have
I got it wrong?). I did add the TUN/TAP option when Ethertap 
got into the deprecated category.

> I found more that needs to be done other than configuring a TAP/TUN 
> device. 

Indeed. The TUN/TAP device must be configured at runtime.
Currently mol runs the script '/usr/lib/mol/bin/tunconfig'
(if it exists). The following config file works for me:

---------------------------------------------------------------
#!/bin/bash
/sbin/ifconfig tun0 192.168.1.1
/sbin/iptables -D POSTROUTING -t nat -s 192.168.1.0/24 -d ! 192.168.1.0/24 -j MASQUERADE >& /dev/null
/sbin/iptables -t nat -s 192.168.1.0/24 -d ! 192.168.1.0/24 -A POSTROUTING -j MASQUERADE
echo 1 > /proc/sys/net/ipv4/ip_forward
---------------------------------------------------------------

The /etc/molrc file should contain 

	netdev:	tun0 -tun

and optionally (for AppleTalk)

	netdev: eth0 -sheep

The MacOS TCP/IP panel should be configured according to

	IP:		192.168.1.2
	Netmask:	255.255.255.0
	Gateway:	192.168.1.1

TUN/TAP as well as iptables and masquerading needs to be on in the
kernel.

> Additionally, the keyboard issue:
> 
> > There is a section about key remappings in the userguide above; does
> > it help? 
> 
> Not really. Why would keystrokes that work under Linux shell *and* work
> under X not work under MOL?

The problem is that MacOS wants to know which physical key is pressed
(in terms of ADB keycodes). In X-mode, MOL gets the following information:

The char bound to the key, say 'n', and the keycode (which is private
to the X-server). Different X-servers use different keycodes 
so relying on it won't work.

So, does the 'n' correspond to the 9:th key in lower row
as is the case on US keyboards? Or is a german keyboard used 
with a german keymapping? In that case the 'n' and 'm' keys 
are swapped and MacOS should be told that the 10:th 
key was pressed (I don't know if it is the n and m keys that
are swapped on German keyboards, but one pair of keys are 
swapped).

>(note: I have not verified any differences
> between fullscreen vs X mode) I have a straightforward no-frills Mac USB
> extended keyboard, US layout preferred. Key Caps under MOL show a layout
> that does not match my keyboard. IIRC, most keys are OK except for `~, ESC,
> one of the shift keys (right one, I think), the arrow keys, and the nav
> (i.e. home/pageup/pagedown etc) keys. I don't know enough to determine
> whether this is a mapping issue, or a layout issue, or if there is a
> difference.

In full-screen mode, MOL receives keycodes (conforming more or
less to the PS/2 keycodes). However, certain ppc distributions 
sometimes use different keycodes for certain keys.

> Would this matter when running MOL in fullscreen mode? This is another
> element it would be nice to have docs on: when does MOL rely on X and when
> does it not? Can we not have MOL reliant on X at all?

Yup. MOL only uses X for the mol-in-a-window mode. The console mode 
is completely independent. Thus, it would be necessary to run the config 
program once in full screen mode and once in X-mode.


Cheers,

/Samuel